Matched amplification and switch joint transform correlator

Abstract

A design of a compact, adaptive, noise robust, and ultra fast correlator is presented. For realizing a correlator with these features the following design steps are taken: For noise robustness a matched amplification-switching is implemented prior to the correlation process. For structure compactness, a lens is integrated in the input plane of the correlator, and a slap of multi devices is integrated in the Fourier plane. The slap consists of a spatial light modulator, a lens, a polarizer, and a light controlled by light modulator for performing matched amplification- switching. These integration procedures allow to built a correlator as small as one cubic cm3. The additional features allow this correlator to perform 105 correlation/sec.

Description

REFERENCE TO RELATED APPLICATION

[0001] This application claims the benefit of provisional application 60/239,836 filed Oct. 12, 2000.

BACKGROUND OF THE INVENTION

[0002] The present invention relates to the field of optical correlator and image clean up and devices. Correlators have been under extensive study for more than a decade, due to their significance in various applications of science and technology. They have been proposed for use in a variety of application such as security, finger print identification and machine vision(H. Rajbenbach, S. Bonn, P. Refregier, P. Joffre, J. P. Huignard .H. S. Jensen and E. Rasumussen, “Compact photorefractive correlator for robotics applications,” Appl. Opt. 31, 5666-5647 (1992) and tracking(Allen Pu, Robert Denkewalter and Demetri Psaltis “Real-time vehicle navigation using holographic memory,” Opt. Eng 10, 2737-2746, (1997). K. Curtis and D. Psaltis, “3-dimensional disk based optical correlator,” Opt. Eng. 33, 4051-4054 (1994).

[0003] So far, to my knowledge, three successful optical correlators have been built for this tracking purpose, The first is the TOPS one TOPS optical correlation progranLindell, Scott D.; AA(Martin Marietta Astronautics Group) Publication: Proc. SPIE Vol. 1958, p. 7-18, Transition of Optical Processors into Systems 1993, David P. Casasent; Ed. The size of this correlator is less than a one cubic foot, and manages to correlate 800 correlation per/ sec. For this correlation the binary phase -only filter was used in the correlation plane. This correlator proved its success in tracking. A more compact correlator was built by Cortec, Inc(11) at Burlington M. A. For this correlator quantum well photorefractive materials with response time of less than Uses time were used . The size of this correlator was the size of a hand and managed to correlate nearly 10,000 correlations per sec. A group in Caltech demonstrated an opto-electronic correlator which can correlate 30,000 correlation per/ sec(K. Curtis and D. Psaltis, “3-dimensional disk based optical correlator,” Opt. Eng. 33, 4051-4054 (1994) This correlation system has been successful in real- time vehicle navigation. This correlation system uses a holographic data base of correlation filter stored on a DuPont BRF-150 photo polymer.

[0004] In accordance of the present invention, a design of a compact optical correlator with operating speed exceeds 1,000,000 correlation/sec is illustrated. Correlator with this massive capability can be used in variety of application involved a large data base for comparing such as finger print identification, information search on the Internet, DNA sequence codes. Templates for machine vision.

BRIEF SUMMARY OF THE INVENTION

[0005] In accordance of the present invention, a design of a compact, adaptive optimal and ultra fast correlator, which may perform more that 105 correlation/sec is presented in FIG. 1. For realizing a correlator with these features the following design steps are taken: (a) For cleaning the input images or the reference templates prior to the correlation process, an adaptive matched amplification-switch device is used (b) For structure compactness, an integrated slap of either matched amplification-switching modulator with a spatial light modulator is presented. (c) For further compactness, the correlator's input device and the integrated slap each is integrated with a transformation lens. This integration allows building a correlator as small as one cubic cm3. (d) For realizing a massive correlation operations, a reference information is read-out of an image multiplixer (preferentially holographic multiplexer which superimposed 10-100 correlation) and then is fed into an image rotator which can rotate the reference information in a time scale of &mgr;sec. This massive information is used to amplify or to switch the input signal prior to the correlation process.

DESCRIPTION OF THE DRAWINGS

[0006] The various features of the invention will become apparent upon study of the following description taken in conjunction with the drawing which:

[0007] FIG. 1: A Schematic diagram of the compact holographic correlator.

[0008] FIG. 2: The architecture of (a) Phorefractive two-beam coupling (b) Matched-amplification with two-beam coupling (c) Optical switching via controlled absorption in semi conducting material.(d) Matched- switching using controlled absorbtion modulator.

[0009] FIG. 3: Architectures which illustrates the structure of (a) The matched amplification JTC and (b) The matched-switch correlator.

[0010] FIG. 4: High contrast optical modulator using Franz-Keldysh effect in thin film of GaAs.

[0011] FIG. 5: Slaps (a) A Slap of a real-time hologram and a spatial light modulator (b) A Slap of controllable absorber- spatial light modulator.

[0012] FIG. 6: The compact structure of the optical correlator

[0013] FIG. 7 The set- up for an image rotator (a) Using Dove prism, (b) Using the components within Dove prism (c) Using an image rotator based on acousto-optic modulators.

DETAILED DESCRIPTION OF THE INVENTION

[0014] FIG. 1 shows a schematic diagram of the ultra fast compact correlation system. In this correlator the reference information (nearly one hundred images all super imposed one on top of the other and propagated in the same direction) is fed to the correlator from an image multiplexer 103. The image multiplexer can be either a holographic storage or a spatial light modulator addressed optically or electronically. The image multiplexer stores various projection, rotation and scale of the target.

[0015] Before the reference information is fed up to the optical correlator, it is fed into an image rotator 113 (See Eung Gi Peak, Joon Y. Choe and Tae. K. Oh, John H. Hong and Tallis Y. Chang. “Nonmechanical image rotation with an acousto-optic dove prism,” Opt. Letts. 15, 1195-1197 (1997 which, consist of four components, 107, 109, 111, 124 discussed later in FIG. 3. This image rotator 113 rotates the multiplexed reference information in &mgr;sec time-scale. After the reference information is rotated then it is fed into a compact matched amplification- switch correlator 123. The out correlation is detected via a CCD 125 or a 2-D imaging detector. The time-scale of the matched amplification-switch is in the range of msec-nsec. The image rotation can be achieved in &mgr;sec time Eung Gi Peak, Joon Y. Choe and Tae. K Oh, John H. Hong and Tallis Y. Chang. “Nonmechanical image rotation with an acousto-optic dove prism,” Opt. Letts. 15, 1195-1197 (1997). Therefore, the number of correlations, which may be achieved in the matched correlation scheme is limited by the time-scale of image rotation. Assuming that one hundred images are fed simultaneously from the input image multiplexer 103 into the optical correlator, then it should be possible to achieve nearly one hundred million correlations per sec (0.1 Ghz correlation rate). In case of matched-amplification the number of correlations can be which, achieved per sec is limited by the matched-amplification process. Therefore, the number of correlations which can achieved per sec (assuming that a hundred images are fed from the spatial light modulator) is 105 (0.1 Mhz).

[0016] Most of the components used in system's design it is well within the skill of the worker in the art.. However, the matched amplification-switch joint transform (MASC) correlation is the hart of the new system's design. With out the MASC design, it would impossible to feed massive information and to process it in real-time. Therefore a special attention is given to the MASC's design and its compact versions. However, in order to understand how the whole system operates it need to explain the operational principle of any of the components in the system.

[0017] In this invention, the operational principles of the following devices are going to be discussed (1) Matched amplification via photorefractive two-beam coupling (2) Matched-switching via InGaAsP controlled absorption absorption modulator (3) The relation between matched-amplification and matched-switching and Wiener filtering. (4) Matched-amplification and matched-switch joint-transform correlator (5) Franz-Keldysh spatial light modulator and their integration with an electro absorption modulator or photorefractive real-time hologram on one slap. (6) Compact design of the matched-amplification or matched-switch joint transform correlator. (7) Image rotator and the possibility of new novel improvement.

[0018] The matched-amplification and the matched-switching are the key component in the new system's design. Both are utilized for cleaning either the input's images or templates form clutter, noise, or unnecessary templates from the multiplexed templates fed into the optical correlator.

[0019] The Matched-amplification device was originally demonstrated proposed by a group at Rockwell, and it was used only to enhance certain features in images (T. Y. Chang , J. Hong and P. Yeh, “Spatial amplification, Opt. Letts, 15, 743-745 (1990). In this device the Fourier transform of a signal image were amplified by a the Fourier transform of a reference image in a photorefractive two beam coupling arrangement.. Here I extend the use of this device for cleaning out of images from their noise and clutter, further more I illustrate how it possible to sit this device to be exactly operating as an adaptive Wiener filter. Applications and operational conditions which were not considered by Chang et. al.

[0020] FIGS. 2(a), and 2(b) illustrate the two-beam coupling and the matched amplification via a photorefractive two-beam coupling. In two-beam coupling FIG. 2(a), when two coherent beams 201 (strong reference beam) and 203 (weak signal beam) interact within a photorefractive crystal 205, they interfere (Introduction to photorefractive nonlinear optics, P. Yeh, Wiely (1993) and generate carriers. These carriers migrate, trapped and generate grating via the Electro-optic effect. This grating is shifted by &pgr;/2 with respect to the interference pattern. The presence of such a quarter-cycle phase shift in the refractive index makes possible a non-reciprocal steady state energy transfer between two beams. At the output the weak signal beam 203 at the input is amplified to produce a strong signal beam 207. The strong reference beam 201 at the input is de-amplified to produce a weak reference beam 209 at the output.

[0021] In the matched amplification instead to amplify a weak beam by a strong beam, the Fourier transformation of one image amplify selectively the Fourier transform of other image in order to achieve image clean up from noise and clutter. FIG. 2(b) illustrate how this scheme operates. Input reference image “A” 211 (in the strong intensity reference beam 213) and a weak intensity noisy signal image templates “AB” 221 (In the weak intensity beam 222) are Fourier transformed by their respective Fourier transform lenses 215 and 223 into a photorefractive crystal 217. The Fourier spectrum of the reference image “A” 211 selectively amplifies the spectrum of the image “A” within the noisy signal image “AB” 221. After this selective spectrum amplification and Fourier transformation by the respective out put Fourier transform lenses 218 and 225, the template image “A” within the noisy signal image templates “AB” 221 becomes intense and clean out of clutter at the out put signal image 220. The image “A” 211 at the input becomes weak at the out put plane 227.

[0022] The matched amplification in photorefractive medium require that the Fourier transform of the reference and the signal images to interfere and write a hologram in the photorefractive medium. This makes the process of matched-amplification limited by the process of wring a hologram in photorefractive medium (msec response-time). For some application this speed is not satisfactory. Therefore, I propose here to replace the matched-amplification by a matched-switch device.

[0023] The architecture of the matched switching is essentially similar to that of the matched amplification. The InGaAsP controlled absorption modulator is the modulator utilized to illustrate the matched switching. Hence, the operational mechanism of this modulator is going to be illustrated first.

[0024] FIG. 2(c) shows a schematic diagram of controlled absorption InGaAsP modulator (K. J. Ebeling, Integrated Opt-electronics. Waveguided Optics Photonics, Springer-verlag, Chapter 12, Opto-electronics modulator, Pages 466-470. W. Kowalsky and K. J. Ebeling. Opt. Letts, Vol 12 1053-1055 (1987). The modulator 240 consists of an epitaxial layer of InGaAsP 237 grown in InP substrate 231. A test It beam of wavelength 1.3 um lies at the band edge of the quaternary layer of In0.73Ga0.27As0.64P0.36 is incident on the controlled absorption modulator. The absorption coefficient of the epitaxial layer 237 at this wavelength is about 6000 cm−1. This beam is partially absorbed in the epitaxial layer but passes thought the InP substrate 231 without further attenuation because of the smaller band gap wavelength of the InP substrate viz .&lgr;g. A shorter wavelength modulated control reference beam 239 with a wavelength &lgr;c is superimposed on the test signal beam 235. It is absorbed in the epitaxial layer and generates excess charge carriers, which cause a change in the transmission of the test signal beam 235. Time variation in the control reference beam 229 is transferred with no distortion to the test beam 235.

[0025] In the matched switching instead of switching a transmitted beam by the a control beam, the Fourier transformation of reference image in the control reference beam selectively switch the Fourier transform of signal image in the transmitted beam.

[0026] FIG. 2 (b) illustrate how this scheme operates. Reference image “A” 241 (in the control reference beam Ic 243) and noisy signal images templates “AB” 255 (In the transmitted signal beam 256) both are Fourier transformed by their respective Fourier transform lenses 245 and 257 into a controlled absorption modulator 267. The Fourier spectrum of the image “A” 241 selectively switch the spectrum of the template image “A” within the signal image templates “AB” 255. After this selective spectrum switching and Fourier transformation by the respective output Fourier transform lenses 245 and 257, the template image “A” within the noisy signal image tempaltes “AB” 255 is what is mostly transmitted as indicated at the out image 253 in the output transmitted beam 251. While the reference image “A” 241 in the control beam is absorbed as indicated in the faint reference image “A” 253 at the out port transmitted beam It 263.

[0027] For the first order approximation, both matched-amplification and matched-switching can be sit to function as Wiener filter. Wiener filter has been used for several decades in retrieving signal from blurred noisy information.

[0028] The matched-amplification is implemented using a real-time hologram, and requires that both coherent reference and signal beams, in contrast matched-switching is implemented using a controlled absorption modulator and doesn't require the beams to have the same wave length neither lenses with same focal lengths. Though both real-time holography and controlled absorption modulators both belong for a one category light controlled by light modulators.

[0029] Both these devices (matched-switch and matched amplification) can be utilized to ultra fast filtering the noise, clutter, none-matching templates in optical correlation systems. Here is illustrated how these two devices can be integrated within a correlation system to produce two new correlation's principle the matched-amplification and matched-switch joint transform correlator (MAJTC and MSJTC).

[0030] FIG. 3(a) shows the proposed scheme for matched-amplification joint transform correlation. In this configuration the signal S 300 which is imbedded in noise, and the reference R 302 signal, are both Fourier transformed via a lens 305 into a slap 317 of a photorefracative crystal 307, polarizer 309, and spatial light modulator 311. The spatial light modulator can be addressed either optically or electronically via an appropriated correlation filter. However, the simplest design of the spatial light modulator is binary spatial modulator. Therefore this should make the binary phase-only filter the simplest correlation filter to use within the system.

[0031] The crystallographic orientation of the crystal is adjusted so that the Fourier spectra of the reference signal R amplifies the Fourier spectra of the contaminated signal beam S. After amplification the output of the crystal at the critical beam ratio m=e&Ggr;L is almost equivalent to the Wiener filtered version of the input. In photorefractive materials of the 44 mm symmetry, the gained component and the back ground component have different polarizations L. J. Cheng and P. Yeh . “Cross-polarization beam coupling in photorefractive GaAs crystal,” Opt. Letts. 12. 705-707 (1987), L.-J. Cheng, G. Gheen, T. H Chao, H. K. Liu, A Partovi, J. Katz and E. M. Garmire, “spatial light modulator by beam coupling in GaAs Crystlas,” Opt. Letts 12, 705-707 (1987), which means that by using a polarizer 309 it is possible to separate them. After separation, the output out of the crystal goes to spatial light modulator 311 built on the same slap, The total output of the of the slap is Fourier transformed via a lens 313 to produce at the out plane 315 which is prefiltered with a matched-amplification filter.

[0032] The matched -switch correlation approach is similar the matched amplification with some modification. The photorefractive crystal 307 of the slap 317 in FIG. 3 (a) is replaced by a control absorption modulator(components 323+324 ) in the slap 333 in FIG. 3 (B). The input beam 301 in FIG. 3(a)is replace with two read out beams 316 and 317 . Each of the beam can be with different wavelength depends on the electro-absorption modulator used in the design, Therefore, in this correlation implementation, the reference image R 318 and the signal image S 320 may be is necessary to scale differently in order to make the Fourier spectra to match each other in the out put. Other alternatives to set the reference image R 318e and the signal image S 320 in different input planes or to replace the lens 305 by synthesized lens 321 with two focal lengths.

[0033] In recent years there has been extensive study in optical holographic storage. A group at Caltech led by D Psaltis has demonstrated that is possible to achieve 3600 correlations per sec. In this scheme, the reference information was addressed from the holographic storage to a binary spatial filter, which was addressed by the signal K. Curtis and D. Psaltis, “3-dimensional disk based optical correlator,” Opt. Eng. 33, 4051-4054 (1994). Here is used similar techniques, but instead of correlation by one single image coming from the holographic storage to the correlation filter. Here is correlated 10 to one hundred images simultaneously, coming from the holographic storage or image multeplixer into the slaps illustrated above. In the first stage matched application is achieved which means only the appropriate reference signal is amplified and the rest is filtered out. The clean signal goes to the spatial light modulator to correlate with the information stored in the correlation filter.

[0034] So far it was described the operational principles of the matched amplifications switch corerlator. The goal is to construct a noise robust ultra fast compact correlator. Both noise robustness and speed were discussed. In this section is illustrated how it is possible to built an integrated correlator. For building the integrated correlator, firstly it is necessary to integrate the matched-amplification or the matched-switch devices with correlation filter device as single one component, further for better performance it is better to separate the amplified component from the back ground component. Because the amplified component has orthogonal polarization compared with original input component. Then it possible to add a polarizer to separated them. Therefore all the mentioned components can be integrated on one single slab. Further compactness in the design the input Fourier transform lens can be integrated in the input plane of the correlator, and the lens of the output (second lens) can be integrated with the slap described above.

[0035] These device can be integrated in numerous methods using various materials. However, due to the relative simplicity in the integration of controlled absorption modulator or photo refractive based materials made from III-IV family, with spatial light modulator made from similar semiconductor family. Here is illustrated an integration based on using Franz-Keldysh effect based spatial light modulator. However, other integration possibilities are also feasible.

[0036] Two forms of integrated slaps are presented: (a) the holographic-spatial light modulator slap, (b) the control absorbtion-spatial light modulator slap. The integration is illustrated with Franz Keldysh Fabrey-Perot spatial light modulator. However, alternative integration with other spatial light modulators or smart integrated structures is also feasible. Before to illustrate the structure of these slaps, first is illustrated Franz-Keldysh effect based spatial light modulator.

[0037] FIG. 4 shows a schematic diagram of Franz-Keldysh effect based spatial light modulator according to reference 41 Parvis Tayebati “High contrast, high reflectivity, optical modulator using the Franz-Keldysh effect in thin film of GaAs,” Appl. Phys. Letts. 63, 2878-2880 (1993). This device consists of thin undoped GaAs 403 grown commercially on an etch-stop Al0.8Ga0.2As layer. A thin silver electrode mirror 411 is spottered on a 25 mm2 square sample. The silver layer (with refractive index of 0.12-6.08, provides a uniform spectral reflectivity and allows efficient heat removal. The above structured layer is epoxied 407 (epoxying) to a sapphire substrate 409. The GaAs substrate and GaAlAs layer are selectively etched and replaced by contacting to the silver electrodes. A thin layer of indium tin oxide 401 (ITO) is spottered on the sample. This layer plays the double role of the top electrode and partial mirror. The device operates that, the large electro-absorption and electorefraction due to the Franz-Keldysh effect can change the transmissivity or the reflectivity

[0038] FIG. 5(a) shows the slap integrating a real-time hologram of Photorefractive GaAs and high contrast optical modulator using the Franz-Keldysh effect in thin film of GaAs: All fabricated on the same substrate. Two intermediating layers between the GaAs 509 and the spatial light modulator r 513 are constructed, the first is a polarizer 507, and the second is an insulating material of SiO20. 505 The function of the polarization layer 507 is to separate the light coming from the reference and the signal beam, because these two components have different polarizations. An optically addressed spatial light modulator using these techniques has already been demonstrated. The insulating layer of SiO20 505 is added here in order to provide an opportunity for applying an external field on the phoreferactive crystal and to achieve high amplification. The Ag electrode 411 In FIG. 4 is replaced by the transparent ITO layer 501

[0039] FIG. 5(b) shows the slap integrating controlled absorption modulator with a Franz-Keldysh effect modulator. This slap has essentially the same structure as the previous slap except that the real-time holographic material of GaAs or InP is replaced by an epitaxial layer of controlled absorption of InGaAsP 522 modulators grown on InP substrate or GaAs. Since the substrate is InP, then the high contrast optical modulator using the Franz-Keldysh effect in a thin film of GaAs can be replaced by one which is fabricated using InP.

[0040] These devices can be integrated further in a compact structure of an optical correlator. The compact structure of an integrated joint-transform correlator is shown in FIG. 6, the first lens 305 and the input spatial light 303 in FIGS. 6(a) and 6(b) are combined on the front surface 602 of a cubic (or orthorhombic) beam splitter 615. Furthermore, the first lens in the correlator (e.g 305 in FIG. 3(a) or 321 in FIG. 3(B) can be replaced in front surface of the cubic (or orthorhombic) beam splitter by a Fresnel lens 603 integrated on the spatial light modulator. On the second surface of the cubic beam splitter, the slap 317 in FIG. 6(a) (The slap of real-time hologram, polarizer, spatial light modulator or the slap 333 in FIG. 3(b) (The slap of controllable absorber- spatial light modulator and polarizer) and the second lens of the correlator 313 and 329 in FIG. 3(a) and FIG. 3(b) are all combined by integrating a reflection Fresnel lens 611 with the slaps. (See also FIGS. 5(a) and FIG. 5(b) ). This correlator can be made less compact via letting the detector array 613 to be a one focal length away from the other surface of the cubic beam splitter (e.g to be in the position 614n).

[0041] For correlation with many images, it is not efficient to feed the input images one by one. The new additional processes in this corelators (matched switching or matched amplification prior to the correlation process) should allows to feed many multiplexed images simultaneously. Feeding many images simultaneous would be extremely difficult or impossible with out the matched amplification or matched -switch.

[0042] The multiplexed information can be fed in either from electronically/optically addressed spatial light modulator, or from holographic storage. The latter approach is relatively faster using the current technologies. Also for feeding the information in, one may need to rotate the multiplexed image in all directions. Even though this step can be done easily through digital, techniques, however recently there were some reports in optical image rotation in time range of uses. Such a speed is impossible to defeat with the current digital technology. Here it is illustrate how to make fast image rotation and how to feed the information from optical holographic storage.

[0043] The common way to achieve image rotation is to use a Dove prism FIG. 7(a). The image rotation is achieved by rotating a Dove prism mechanically. In FIG. 7(a) the image “P” 701 is images via a lens 703 in to a rotating Dove prism 705. The light, which emerges of the lens 707 is images into the output to produce the rotated version of the input image “P” 709. Chiou and Yeh (40) demonstrated image rotation using Dove prism.

[0044] Other groups demonstrate that is possible to perform image rotation using two acousto-optic detector and cylindrical mirror. This approach is essentially based on the analysis of the conventional Dove prism. As shown in FIG. 7(a) the Dove prism can be divided into three parts shown in FIG. 7(b): the first part is a wedge prism 714 to direct the incident beam to the bottom (or the top) of the second part 716, which functions as a mirror. The third part 718 is another wedge prism to align the beam along the original direction. All these parts rotate together.

[0045] FIG. 7(C) shows the acousto-optic approach for image rotation. In this approach each of the first 714 and the second prism 718 in FIG. 7(b) are replaced by two pairs of cascaded acousto optic deflectors. One acousto optic deflectors for the X direction and the other for the Y direction, The firs pair is shown by the component 728 and the second is shown by the component 732 in FIG. 7 (C). The reflecting mirror, the second part of Dove prism 716 in FIG. 7(b) is replaced by a cylindrical mirror 730 In FIG. 7(c). However, to prevent unwanted distortion owing to the curvature of the circular mirror surface, the cylindrical mirror is discreteized to multiple facets. FIG. 7(c) illustrates this scheme

[0046] This image rotator can be improved in terms of operational speed or Image quality. For enhancing the speed the acousto-optic modulator can be replaced with Electro-optic deflector. For improving the image quality the acousto optic modulator can be replaced with beam steering devices based on liquids crystals.

[0047] For feeding the information in, in the last decade It has been demonstrated that it possible to store and multiplex up to 10000 images in one hologram. These image multiplexers can be used to store the correlator's reference templates. Two forms of image multiplexer are suitable for fast feed in the information (shift multiplexing )Demetri Psaltis, Michael Levene, Allen Pu and George Barbastathis, Kevin Curtis “Holographic storage using shift multiplexing,” Opt. Letts ,20, 782-784( 1995 and Peristrophic multiplexing) K. Curtis, A. Pu and D Psaltis, “Method for holographic storage using peristrophic multiplexing,” Opt. Letts. 19, 993-994( 1994). However for a reconfigurable system it is better to use images multiplexed on spatial light modulator.

[0048] Matched amplification at beam ratio=exp (&Ggr;l) is essentially a Wiener filter (&Ggr; is the coupling coefficient of the material L is the crystal thickness). In order to prove this, let as assume that we have the configuration of two beam coupling in which in the reference beam A1 we have the signal, and in the object beam (beam A2) we have the signal which is imbedded in noise.

[0049] In this configuration both the object and the reference signal are Fourier transformed via a lens into a photorefractive crystal, the crystal is oriented in a such a way that the energy is transferred from the Fourier transom of the signal to the Fourier transom of the reference beam.

[0050] Then the output from the crystal can be written as: 1 output = A 2 ⁡ ( S + N ) ⁢ 1 + mX 2 1 + mbX 2

[0051] Where X is defined as 2 x = S S + N

[0052] What is inside the square root can be written in the following form 3 1 + mx 2 1 + mbx 2 = 1 b ⁡ [ 1 + ( b - 1 ) ⁢ 1 1 + mbx 2 ] = [ 1 + 1 2 ⁢ ( 1 b - 1 ) ⁢ mbx 2 1 + mbx 2 - 1 8 ⁢ ( 1 b - 1 ) ⁢ mbx 2 1 + mbx 2 ] 1 2

[0053] For &Ggr;L Positive 1\b-1<1, therefore the above term can be expanded to be 4 1 + 1 2 ⁢ ( 1 b - 1 ) ⁢ mbx 2 1 + mbx 2 - 1 8 ⁢ ( 1 b - 1 ) 2 ⁢ ( mbx 2 1 + mbx 2 ) 2

[0054] In photorefractive materials of the 4 mm symmetry, the first term, and the high order terms have different polarizations, which make it possible to separate the first terms from the higher terms.

[0055] Let assume that a separation from the first order and the higher orders has been achieved, and the contribution of the third order to the second order is negligible, then the output from the crystal can be written as 5 A out = A 2 ⁡ ( S + N ) ⁢ ( &LeftBracketingBar; S &RightBracketingBar; &LeftBracketingBar; S + N &RightBracketingBar; ) 2 1 + mb ⁡ ( &LeftBracketingBar; S &RightBracketingBar; &LeftBracketingBar; S + N &RightBracketingBar; ) 2 = m ⁢   ⁢ A 2 ⁡ ( S + N ) ⁢ &LeftBracketingBar; S &RightBracketingBar; 2 &LeftBracketingBar; S + N &RightBracketingBar; 2 + mb ⁢ &LeftBracketingBar; S &RightBracketingBar; 2

[0056] For a signal that is highly contaminated within a noise, it is possible to approximate the above equation in the following form: 6 A out = m ⁢   ⁢ A 2 ⁡ ( S + N ) ⁢ &LeftBracketingBar; S &RightBracketingBar; 2 &LeftBracketingBar; N &RightBracketingBar; 2 + m ⁢   ⁢ b ⁢ &LeftBracketingBar; S &RightBracketingBar; 2

[0057] At the critical beam intensity ratio m=e&Ggr;L, the output from the crystal becomes, 7 A 2 ⁢ e Γ ⁢   ⁢ L ⁢ ( S + N ) ⁢ &LeftBracketingBar; S &RightBracketingBar; 2 &LeftBracketingBar; S &RightBracketingBar; 2 + &LeftBracketingBar; N &RightBracketingBar; 2

[0058] In similar manner it is possible to prove also that matched switching using controlled absorption spatial light modulator is also equivalent to the wiener filter.

[0059] In controlled absorption modulator, the change in the transitivity of the transmitted beam as a result of adding the control beam Ic is given by: 8 Δ ⁢   ⁢ t t = t ⁢ ( I c ) - t ⁢ ( I c = 0 ) t ⁢ ( I c = 0 ) = α ⁢ ( λ t ) ⁢ d ⁢ ⟨ Δ ⁢   ⁢ n ⟩ n 0 + ⟨ Δ ⁢   ⁢ n ⟩ + N c ⁢ exp ⁢ { h ⁢   ⁢ c / λ t - W g k ⁢   ⁢ T }

[0060] Where <n>=Ic&tgr;/Ah &ohgr;sd, &tgr;s is the charge carrier response time, h is blank coefficient, K is Boltzman constant, Wg is the energy gap of the material

[0061] Let as assume that in the configuration of controlled absorption InGaAsP modulators that the reference beam ( the control beam) Ic has the signal, and in the object beam (beam A2) It has the signal which is imbedded in a noise.. Also let us assume that the object in the reference beam is scaled adjusted, so the Fourier order in the reference and the signal beam correspond to the same special frequencies. Since <&Dgr;n>∝|S|2, then the relative change in the transmissivity is given by 9 Δ ⁢   ⁢ t t = α ⁢ ( λ t ) ⁢ d ⁢   ⁢ w 1 ⁢ &LeftBracketingBar; S &RightBracketingBar; 2 w 1 ⁢ &LeftBracketingBar; S &RightBracketingBar; 2 + n 0 + N c ⁢ exp ⁢ { h ⁢   ⁢ c / λ t - ω g k ⁢   ⁢ T }

[0062] w1 is proportionality weighing factor depends on the wavelength and the material characteristics

[0063] Since either no and Nc are homogeneously distributed on the sample, and have no dependence on the intensity of the signal beam, then, for additive Gaussian noise, both no and Nc are proportional to the distribution of the additive noise. For signals imbedded in a white additive noise, it is possible to prove that 10 Δ ⁢   ⁢ t t = α ⁢ ( λ t ) ⁢ d ⁢   ⁢ w 1 ⁢ &LeftBracketingBar; S &RightBracketingBar; 2 w 1 ⁢ &LeftBracketingBar; S &RightBracketingBar; 2 + w 2 ⁢ ⟨ N ⟩ 2

[0064] Where w1 and w2 are weighing factors. It is possible to adjust the input beam ratio so that w1=w2, and hence the change in the transmissivity becomes equivalent to Wiener filtering.

Claims

1. A matched amplification-switching optical image processor, comprising:

(a) A beam of coherent light for carrying an cluttered input image;

(b) A lens for receiving the beam and performing a spatial Fourier transform of the cluttered input image;

(c) A light controlled by light modulator at the Fourier plane for receiving the transformed clutter image;

(d) A beam of coherent light for carrying a reference pump input image;

(e) A lens for receiving the beam and performing a spatial Fourier transform of the reference input image;

(f) A means for causing the coherent reference pump image to illuminate the matching spatial frequency components of the transformed cluttered image at the Fourier plane, thereby leaving the clutter's spatial frequency components unilluminated and energy transmitivity of the clutterd image is enhanced only in the matched illuminated areas of the light controlled by light modulator;

(g) a second lens for receiving the transformed cluttered image from the Fourier plane and performing a spatial Fourier transform, thereby reproducing nearly a cluttered free image at the output;

2. a matched amplification-switching image processor of claim 1, further comprising a Fourier transform mixing mean coupled between light controlled by light modulator and a second Fourier transform lens for mixing a matched switch-enhanced information with a processed form of the Fourier transformation of the other input information;

3. Matched amplification-switching image processor mean, of claim 1 wherein said a light controlled by light modulator is selected from the group consisting of:

(a) A controlled absorption modulator;

(b) A real-time holographic two-beam coupling modulator;

4. Optimal operational conditions of the image processor via matched amplification-switching mean of claim 1, comprising:

(a) Siting the beam ratio to be equal to the exponential gain(e.g m=exp(gamma &Ggr;L) for the real-time holographic two-beam coupling modulator;

(b) Siting that w1=w2 Where w1 and w2 are weighing factors depends on the beam ratio.

5. Matched amplification switch- optical correlation apparatus comprising:

(a) Input reference images multiplexing mean;

(b) A beam of coherent light for carrying a cluttered input image from an input detection system;

(c) A beam of coherent light for carrying multiplex input template from reference image multiplexing mean;

(e) A transformation means for transformation of image multiplex information;

(f) A matched switch-amplification mean for cleaning the clutter and the none matched templates out of the cluttered input image and the multiplexed templates;

(g) A transformation mixing mean for mixing a matched amplified-switch information at light controlled by light modulator with a processed form of the Fourier transformation of other input information;

(h) A beam of coherent light for carrying the mixed information;

(i) A transformation means for transformation the mixed information to output detector to produce a matched- switch amplification correlation;

6. Slap for matched amplification-switching processing mean comprising:

(a) A Light controlled by light modulator; and

(b) Spatial light modulator;

7. The slap of claim 6, further comprising an optical polarizer for separating the matched switched-amplified component from the back ground component

8. Compact cubic correlator structure comprising:

(a) An integrated lens on the first surface of a cubic beam splitter;

(b) Integration of a lens, controlled light by light modulator, polarizer and lens on the second surface of the cube beam splitter;

(c) An integration array of detector in an tranformation distance from the second surface of the cube;

9. A System with a massive correlation capability comprising:

(a) An image multiplexing mean for multiplexing reference templates;

(b) A beam of coherent light for carrying multiplex input template from image multiplexing mean to the input plane of matched switch-amplification correlator;

(c) A Detection mean for detecting the cluttered input image;

(d) A beam of coherent light for carrying an cluttered input image to the input plane of matched switch-amplification correlator;

(e) A matched amplification-switch optical correlator mean;

(10) The system of claim 9, further comprising an image rotation mean for rotating the muliplexed images;